## Key Takeaways

- **Legal busywork is the real bottleneck.** Intake triage, contract tracking, NDA turnaround, and status chasing eat the hours that should go to actual legal judgment.
- **A review-first AI employee fits legal better than a silent agent.** Viktor drafts the work and waits for a lawyer to approve before anything is sent, so the judgment never leaves the human.
- **It runs from where legal already gets pinged.** Most legal requests start in Slack. Viktor lives there, so intake, triage, and status updates happen in the existing flow.
- **It connects to the real stack.** Viktor reaches 3,200+ tools, so it can track agreements in SignWell or DocuSign, log matters in Linear or Notion, and pull context from Gmail.
- **The line is firm: drafting and tracking, not deciding.** Viktor handles the gathering and first drafts. A lawyer owns every legal call.
- **The payoff is response time.** Faster intake and turnaround on routine requests, without adding headcount or cutting corners on review.

In-house legal almost never drowns because the legal questions are hard. It drowns in the connective tissue around them: the intake requests scattered across Slack and email, the NDAs that need a quick turnaround, the contracts whose renewal dates live in someone's memory, the endless "what is the status of my thing?" pings. None of that is legal work. All of it routes through the legal team and crowds out the work only a lawyer can do.

This post is about handing that connective tissue to an AI employee while keeping every legal decision firmly with a person. We will be specific about what to delegate, what to never delegate, and how the review model keeps a small legal team in control.

## Where does a legal team's time actually go?

Walk through a typical week on a small in-house legal team and the same kinds of requests pile up:

- a sales rep drops an NDA request in Slack
- a vendor contract needs a routine review
- someone asks whether the MSA with a customer auto-renews next month
- a new hire needs a standard agreement
- marketing wants a quick claims check before a launch

Individually, each is minutes. Together, they are the job. The high-judgment work, the negotiation, the genuinely novel risk question, the strategic call, gets whatever attention is left after the queue is cleared. This is the same dynamic [Slack's Workforce Index measured](https://slack.com/blog/news/the-workforce-index-june-2024) across desk workers, who report spending about a third of their day on tasks they consider low-value, and legal feels it acutely because the routine requests are constant and the senders all think theirs is urgent.

[Stanford's 2024 AI Index](https://aiindex.stanford.edu/report/) noted a 32 percent year-over-year rise in publicly reported AI incidents, a reminder that the right model for legal is not "let an agent run free." It is to remove the busywork while keeping a human firmly on every judgment. That distinction is the whole design.

## What a legal team can safely hand off

The safe-to-delegate work is the gathering, drafting, tracking, and chasing. None of it is the legal decision itself. Here is where an AI employee earns its place, and where it must stop.

### Intake and triage

Legal requests arrive everywhere. Viktor can sit on the intake channel, capture each request, ask the standard clarifying questions, tag it by type and urgency, and log it so nothing falls through. The lawyer opens a clean, sorted queue instead of digging through scattered threads.

### Contract and renewal tracking

The renewal date nobody remembers is a recurring source of pain. Viktor can keep a register of agreements in Notion or a spreadsheet, watch signature status in SignWell or DocuSign, and flag renewals and expirations before they become a fire drill.

```prompt
@Viktor every Monday, check our contract tracker in Notion and SignWell.
List any agreement that renews or expires in the next 45 days, plus anything
still awaiting signature for more than 7 days. Post it in #legal and tag me.
```

### First-draft routine documents

For standardized documents like NDAs from an approved template, Viktor can prepare a first draft populated with the request details, then hand it to a lawyer to review and approve before anything goes out. The lawyer edits or signs off. The AI employee never sends on its own.

### Status updates and chasing

The "where is my thing?" pings are pure overhead. Viktor can answer status questions from the tracker and nudge the right person when a signature or an internal review is overdue, so the legal team stops being a help desk for its own queue.

## What a legal team should never hand off

This is the part that matters most, so it gets its own section. An AI employee does not make legal judgments. It does not decide whether a clause is acceptable, whether risk is tolerable, whether to approve or reject terms, or what advice to give. It drafts, tracks, and gathers. A lawyer decides.

Viktor is review-first by default, which makes this boundary structural rather than a matter of good intentions. For anything that changes a record or goes outside the team, Viktor drafts and waits for explicit human approval. So a generated NDA sits as a draft until a lawyer signs off. A renewal flag is a heads-up, not an automatic action. The judgment cannot leak out, because nothing leaves without a person saying yes. We explain why that default matters for any high-stakes function in [Don't let your AI agent act without asking](https://viktor.com/blog/dont-let-ai-agent-act-without-asking).

## Delegate versus decide: a clear split

| Task | AI employee (Viktor) | Lawyer |
| --- | --- | --- |
| Capture and tag an incoming request | Yes, from the Slack intake channel | Reviews the sorted queue |
| Track renewal and expiration dates | Yes, flags ahead of deadline | Decides whether to renew |
| Draft a standard NDA from a template | Yes, as a reviewable draft | Approves or edits before it sends |
| Decide if a contract clause is acceptable | No | Yes, always |
| Chase an overdue signature | Yes, sends the nudge on approval | Sets the policy |
| Assess legal risk or give advice | No | Yes, always |
| Assemble context for a review | Yes, pulls the relevant docs | Makes the call |

The left column is hours of recurring overhead. The right column is the work you hired a lawyer to do. Drawing the line this cleanly is what makes delegation safe for a function where mistakes are costly.

## Getting started without overreaching

Start narrow. Pick the one routine flow that generates the most noise, usually intake or NDA turnaround, and let Viktor own just the gathering, drafting, and tracking around it. Keep every output review-first, so the legal team approves before anything leaves. Once the queue is calmer and the drafts are reliably good, expand to renewal tracking and status chasing.

The goal is not a robot lawyer. It is a legal team that spends its hours on judgment instead of logistics. For the broader pattern of handing recurring operational work to a coworker, see [AI for operations teams](https://viktor.com/blog/ai-for-operations-teams), and for the management mindset, [How to manage an AI coworker](https://viktor.com/blog/how-to-manage-an-ai-coworker).

## Frequently Asked Questions

### Can AI replace a lawyer on my team?

No. An AI employee like Viktor handles the busywork around legal work: intake, tracking, first drafts of standard documents, and status chasing. It does not make legal judgments, assess risk, or give advice. A lawyer owns every legal decision, and the review-first model keeps it that way.

### What legal tasks is it safe to delegate to an AI employee?

The gathering, drafting, and tracking. That includes capturing and tagging intake requests, tracking renewal and expiration dates, preparing first drafts of standardized documents like NDAs for human review, and chasing overdue signatures. Anything that requires a legal judgment stays with a lawyer.

### How does Viktor keep legal work under control?

Viktor is review-first by default. For anything that changes a record or goes outside the team, it drafts the work and waits for explicit human approval before acting. A generated NDA stays a draft until a lawyer signs off, and a renewal flag is a heads-up, not an automatic action.

### Which tools can it connect to for legal work?

Viktor connects to 3,200+ tools. For legal teams that commonly means tracking signatures in SignWell or DocuSign, logging matters in Linear or Notion, and pulling context from Gmail, all triggered from Slack where most legal requests already start.

### Where does the legal team interact with Viktor?

In Slack or Microsoft Teams. Because most legal requests already arrive there, Viktor handles intake, triage, and status updates inside the existing flow. There is no separate legal portal to maintain or new app for requesters to learn.

### What is a realistic first use case for a small legal team?

Intake and triage, or NDA turnaround. Pick the routine flow that generates the most pings, let Viktor capture, sort, and draft around it, and keep every output review-first. Once that is reliable, expand into renewal tracking and overdue-signature chasing.

---

**Viktor is an AI employee that lives in Slack, connects to 3,000+ integrations, and does real work for your team.** [Add Viktor to your workspace -- free to start →](https://viktor.com/?utm_source=blog&utm_medium=cta&utm_campaign=ai-for-legal-teams)